It is the question every SEO asks the moment they take AI Overviews seriously: do I actually need to rank in the top 10 to get cited, or is that just the old SERP thinking carried over. The honest answer, backed by large-scale data rather than opinion, is reassuring and a little surprising. No, you do not need to be in the top 10, though it still helps. Here is exactly what the numbers say, why a top-10 spot is no longer a gate, and what to do whether you rank there or not.
The short answer
You do not need to be in the top 10 to be cited in a Google AI Overview, but ranking well still improves your odds. The most recent large study found that only about 38 percent of AI Overview citations come from pages ranking in the top 10, down sharply from roughly 76 percent a year earlier, with the rest split between positions 11 to 100 and beyond position 100. So a top-10 position is the single strongest lever, but it is now a minority of citations. Because Google breaks a query into sub-questions and pulls answers from across the web, a page that ranks modestly for the head term can still be cited if it answers a related question better than anyone else.
What the data actually says
Let us put real numbers on it, because this question attracts a lot of guesswork. Ahrefs analyzed 863,000 keyword SERPs and four million AI Overview URLs and found that 38 percent of cited pages also rank in the top 10 for the query, down from about 76 percent in their July 2025 study. The remaining citations split almost evenly, roughly 31 percent from positions 11 to 100 and roughly 31 percent from beyond position 100. Read that again: nearly two thirds of AI Overview citations now come from pages that are not in the top 10 for the query being answered. A year earlier, the top 10 dominated. The trend is unmistakable, citations are spreading down and across the results, not concentrating at the top.
Here is the shift at a glance.
| Where the cited page ranks for the query | Share of AI Overview citations (2026) | Share a year earlier |
|---|---|---|
| Top 10 | About 38 percent | About 76 percent |
| Positions 11 to 100 | About 31 percent | The minority |
| Beyond position 100 | About 31 percent | The minority |
The headline is that first row falling from 76 to 38. If your whole AI Overview strategy assumed you must rank top 10 first, the data says you have been working with an outdated map.
Why top 10 is not a gate: query fan-out
The mechanism behind the spread is query fan-out, and understanding it changes how you work. When you ask Google something, the system does not just answer that one query. It breaks it into several related sub-questions, gathers candidate passages that answer each of them, and synthesizes one response with supporting links. So the page that gets cited is not necessarily the one ranking first for your exact head term, it is the page that best answers one of the sub-questions in the cluster. A modestly-ranking page that nails a specific sub-question can be pulled in alongside, or instead of, the top-ranked page for the broad term. That is why breadth across a question cluster now matters as much as a single ranked position, a point I cover in depth in Google AI Overview ranking factors.
So does ranking matter at all?
Yes, do not overcorrect. A top-10 position is still the strongest single predictor of being cited, just not a requirement. If you rank first, your odds of citation are meaningfully higher than if you rank fortieth, all else equal. The right way to hold both facts at once is this: ranking well is a strong tailwind, not a turnstile. You should still pursue strong rankings on your priority queries, because they help. You should simply stop treating a top-10 spot as a prerequisite before you bother optimizing for citation, because that belief leaves most of the opportunity, the nearly two thirds of citations outside the top 10, on the table.
What matters as much as ranking
If position is only part of the story, what fills the rest. Three things. First, coverage of the question cluster: because of fan-out, the more of the related sub-questions you answer well, the more chances you have to be cited, even on pages that do not rank first. Second, extractability: AI Overviews quote passages, so a clean, direct answer near the top of each section gives the model something liftable, which a buried answer does not. Third, being genuinely discussed: Ahrefs studied 75,000 brands and found that branded web mentions and YouTube mentions correlate far more strongly with AI visibility than backlinks or domain rating, so being talked about lifts your presence independent of where any single page ranks. None of these require a top-10 spot. All of them are within reach of a focused, mid-ranking site.
What to do if you are not in the top 10
This is the practical part, because most pages are not top 10 for their hardest terms. Pursue the sub-questions, not just the head term. Identify the cluster of related questions around your topic and answer each one thoroughly on its own well-structured page or section, since fan-out can cite you for those even when you do not lead the main query. Make every answer extractable, a direct two to three sentence response under a clear question heading. Confirm technical eligibility, because Google states a page must be indexed and eligible for a snippet to appear in AI features, and losing snippet eligibility removes you regardless of rank. And keep earning mentions, since being discussed raises your odds across the board. In short, you compete on answering specific questions better, not on outranking incumbents for one broad term.
A worked example
A mid-size site ranked around position 15 to 30 for its main category term and assumed it was therefore invisible in AI Overviews. When it actually checked, it found it was already being cited for several narrower sub-questions where its content was unusually clear and specific, exactly the fan-out effect. Rather than chasing the brutal head term, it leaned into the pattern: it built out thorough, well-structured answers to the cluster of related questions, made each answer liftable, and earned a few genuine mentions. Its AI Overview citations grew, none of which required cracking the top 10 for the head term. The lesson it took away was that the top-10 obsession had been hiding the opportunity sitting in the long tail of sub-questions.
Common misconceptions
A few beliefs keep people stuck. The first is that you must rank top 10 first, when the data shows nearly two thirds of citations come from outside it. The second is the opposite overcorrection, that ranking no longer matters, when it remains the strongest single signal. The third is that AI Overview citation is purely about one page versus one query, when fan-out means it is about answering a whole cluster. The fourth is that since clicks are falling anyway, citation is pointless, when in fact, because click-through roughly halves once an AI Overview appears, being the cited, named source is increasingly the visibility itself. The balanced view holds all four corrections at once.
How to find your opportunities
The fastest way to act on this is to map the sub-questions you can realistically win, rather than fixating on head terms where you do not rank. I use SQSEO for this, fanning one seed keyword into the longtail and question-level prompts that trigger AI answers, for free, so you can see the cluster of related questions and where you already have a citable angle even without a top-10 position. Then you answer those questions thoroughly and check whether you start getting cited. It turns the vague worry of am I ranking high enough into a concrete list of winnable sub-questions, which is the whole shift this data demands. The deeper question of whether authority metrics matter, using Perplexity as the example, is in are Perplexity answers based on domain rating.
A note for new and small sites
This data is especially encouraging if you are newer or smaller and cannot realistically crack the top 10 for competitive head terms yet. In the old model, that locked you out: no top-10 ranking, no visibility. Fan-out changes the math. Because nearly two thirds of AI Overview citations come from outside the top 10, a focused page that answers a specific sub-question better and more clearly than larger competitors can be cited next to them, even while your domain is still climbing on the broad terms. The strategic move for a smaller site is therefore to go narrow and deep: pick the specific questions in your niche where you can genuinely write the best, clearest answer, and own those, rather than throwing yourself at head terms where incumbents are entrenched. AI Overviews reward the best answer to a specific question, and that is a fight a small, sharp site can actually win.
How to check if you are already cited
Before assuming you are invisible, look. Ask Google your category questions with AI features on and read which pages get cited, then do the same for the narrower sub-questions around each topic. Many sites discover, as in the example above, that they are already being cited for specific sub-questions while losing the head term entirely, which is the fan-out effect working in their favor unnoticed. Keep a simple log of which questions cite you, which cite competitors, and which cite nobody useful, because that grid tells you exactly where to double down and where the open opportunities are. Checking takes minutes and replaces the anxious guess of am I ranking high enough with concrete evidence of where you already appear and where you can.
The bottom line
No, you do not need to be in the top 10 to rank in AI Overviews. A top-10 position helps and remains the strongest single signal, but it now accounts for only about 38 percent of citations, down from 76 percent a year earlier, with the majority coming from pages ranking lower thanks to query fan-out. So stop gating your AI Overview strategy on a top-10 spot. Compete by answering the whole cluster of sub-questions clearly and extractably, stay eligible, and keep earning mentions. That is a game a mid-ranking site can win, which is exactly why the falling top-10 share is good news, not bad.
Frequently asked questions
Do I need to be in the top 10 to rank in AI Overviews?
No. A top-10 position helps and is the strongest single signal, but only about 38 percent of AI Overview citations come from top-10 pages, down from roughly 76 percent a year earlier. The majority now come from pages ranking lower, because Google fans a query into sub-questions and cites pages that answer those well regardless of their rank for the head term.
Why are pages outside the top 10 getting cited?
Because of query fan-out. Google breaks your query into related sub-questions, gathers passages that answer each, and synthesizes one response. A page that ranks modestly for the main term can be cited if it answers a sub-question better than anyone else, which spreads citations across positions 11 to 100 and beyond.
Does ranking still matter for AI Overviews?
Yes. Ranking well remains the strongest single predictor of being cited, it is just no longer required. Treat a strong ranking as a powerful tailwind rather than a prerequisite, and pursue it alongside cluster coverage, extractable answers, and earned mentions rather than waiting for a top-10 spot before optimizing.
What should I do if I am not in the top 10?
Compete on the sub-questions. Map the cluster of related questions around your topic, answer each thoroughly with a direct, liftable answer near the top, confirm your pages are indexed and snippet-eligible, and keep earning mentions. Fan-out can cite you for those sub-questions even when you do not lead the main query.
